Effects of geomorphometric variables variability due to different computing algorithms in environmental modeling

Abstract

Various environmental modeling models use geomorphometric variables in their structure, for estimating environmental variables. These geomorphometric variables have several computing algorithms implementations, which give different values of output for every algorithm. Although there are several papers concerning the ranking of the algorithm accuracy compared with reference data, not always the best ranked algorithm will be used for computing that geomorfometric variable, because of several reasons, including algorithm availability in GIS software. We use gridded mathematical surfaces as source of geomorphometric variables and different algorithms to compute them, in order to test the effects of their value variability in several types of environmental models. Error propagation equations are used as estimators of error due to variables variability, and the variance of obtained value with reference to the average value of the total obtained values using all the algorithms, is used to assess the variability. Because not always a single algorithm can be considered the best estimator, we consider the use of the all possible algorithms and the averaging of the results, which can be a general and a spatial good performance operation for minimizing the errors. © 2010 Published by Elsevier Ltd.